Desert Island XI - QUARANTINE ISLAND - Master List and Discussion Thread

The friendliest place on the web for anyone that follows U2.
If you have answers, please help by responding to the unanswered posts.
I feel bad for neglecting this, but I've been working on it yesterday and today. I'd mentioned that there were two lists I hadn't listened to yet, and bunch that I'd listened to weeks ago but hadn't posted commentary on yet. I've gone through the latter group and refreshed me memory and typed my thoughts out, so I'm about to post those en masse.

Once that's done, I just have Ashley and Cobbler's lists left to listen to. I'm going to try to power through one tonight and the other tomorrow, and then I'll get my rankings in.

This is so like me. When I was in school, I'd always wait until the last minute to do everything.
 
I need to get analytics to influence all of your votes.

Are you going to have time to post any of these before LM's closing date? I would value seeing them. I think I have a ranking settled, but a couple of decisions are really tight and, who knows, maybe some data might swing it. And whether it influences my decisions or not, I'd just be really interested.

I don't say this to apply any pressure though - it could also be fun to have analytics as part of a results thread.
 
Are you going to have time to post any of these before LM's closing date? I would value seeing them. I think I have a ranking settled, but a couple of decisions are really tight and, who knows, maybe some data might swing it. And whether it influences my decisions or not, I'd just be really interested.

I don't say this to apply any pressure though - it could also be fun to have analytics as part of a results thread.



Literally working on it all day today! Excited to show people some stuff. Hopefully will be posted in the next three hours.
 
Fun Data Just for You!

It's here... consider this my apology for falling off the face of the Earth. The data is interactive and designed for you to play with. Some notes:
- use the legend to enable and disable users (some views default to one user, others show all, but you can show whoever you want)
- hover over some of the material for more info (user, artist, song name)
- ask questions if you're bored

https://dixi-hosting.bitbucket.io/

P.S. if people want superlatives, there's obviously a million ways we can spin the data.
P.P.S. I'm not responsible for typos or incorrect data
P.P.P.S. consider this as a beta for what we could make a thing for the next go round ;)
 
Last edited:
Holy shit that is amazing.

The artist popularity one cracks me up, as it's basically "Axver's list is shit nobody else likes".

Love gump's silent track repeatedly showing up at zero. The "loudness" one is especially good for this. :lol:
 
Hey all. I’ll send you my rankings tonight, LM. I’ve been sick for a few days and couldn’t give this too much thought). Looking forward to the results though.
 
Holy shit that is amazing.

The artist popularity one cracks me up, as it's basically "Axver's list is shit nobody else likes".

Love gump's silent track repeatedly showing up at zero. The "loudness" one is especially good for this. :lol:



For the record you can drag the axes to filter. The one thing you can’t do (and I can still code this in) is re-bin the histogram. I defaulted them all to 0.01 if they were 0-1 scaled. Technically I can enable user defined binning.

Thank open sources.
 
It's here... consider this my apology for falling off the face of the Earth. The data is interactive and designed for you to play with. Some notes:
- use the legend to enable and disable users (some views default to one user, others show all, but you can show whoever you want)
- hover over some of the material for more info (user, artist, song name)
- ask questions if you're bored

https://dixi-hosting.bitbucket.io/

P.S. if people want superlatives, there's obviously a million ways we can spin the data.
P.P.S. I'm not responsible for typos or incorrect data
P.P.P.S. consider this as a beta for what we could make a thing for the next go round ;)

Am I reading the Track Energy one correctly? Did EVERY list start slow and the build towards the end ?
 
Hey all. I’ll send you my rankings tonight, LM. I’ve been sick for a few days and couldn’t give this too much thought). Looking forward to the results though.
:up: Hope you're feeling better.

Once you and LN7 have your rankings in, we can wrap this up.
 
Am I reading the Track Energy one correctly? Did EVERY list start slow and the build towards the end ?



Really hard to say for certain. I should put a group index on there (normalize everyone’s start to finish times and then calculate the average energy at each percentage of completion).

Even though DIXI is wrapping up, I want to definitely encourage this to come in full form if we wanna have another go round... I worked on getting data into your hands for your own exploration this time, with little analysis myself... with that being said I need to hold my shit together a little better with actually participating in commentary.
 
Alright, just sent my results to LM.

I gotta say, this was really hard to rank. I thought the quality of the competition was truly exceptional, and I basically ended with two large top tiers that on any day I could have ranked in a different order.
 
Just eyeballing it, LN7 is probably going to decide the winner of this competition. Which is especially interesting because he didn't review anything past group 1 so I have no idea what he thinks. [emoji38]

The top 2 are really, really close.
 
Well I listened to everyone so don’t worry hahaha. And also keep in mind that I’ve gotten intimate with every track list on a totally different level what with data, so have no fear!
 
The lists are all in and we have a winner. Tonight and tomorrow afternoon I will go through LN7's list and pick out some stats to highlight for each participant so that the evening's results thread will have a little extra spice.

Thanks to everyone for being punctual.
 
LN7 advanced stats that will be included in the results thread:

Acousticness average
Artist popularity average
Countries
Danceability average
Energy average
Happiness average
Loudness average
Song popularity average

Gonna get it started at 6 PM PST.
 
Fucking awesome! Thanks LM! My trifecta prediction, in no order, is Gump, LJT and LM. Maybe iYup too.

LN7 can you explain a few things to me.

Historgams - what am I looking at? The lines/graphs don't mention any artists/songs
Acousticness - songs that are most acoustic guitar?
Danceability - what is this
Energy - what is this
Instrumentalness - longest amount of time in songs without words?
Keys - what is this
Liveness - what is this
Loudness - is this a scale of like Exit on the original JT CD to ABOY on Hutdab?
Song mode - what's this
Speechiness - most spoken word/close to spoken word?
Valence/happiness - what the fuck is this
 
Fucking awesome! Thanks LM! My trifecta prediction, in no order, is Gump, LJT and LM. Maybe iYup too.

LN7 can you explain a few things to me.

Historgams - what am I looking at? The lines/graphs don't mention any artists/songs
Acousticness - songs that are most acoustic guitar?
Danceability - what is this
Energy - what is this
Instrumentalness - longest amount of time in songs without words?
Keys - what is this
Liveness - what is this
Loudness - is this a scale of like Exit on the original JT CD to ABOY on Hutdab?
Song mode - what's this
Speechiness - most spoken word/close to spoken word?
Valence/happiness - what the fuck is this



For all charts, you can toggle them by individual user (one color) or multiple users by clicking their names. You can scale each axis by dragging it. For convenience, you and add or remove all users. For some charts (like charts) you can change them to averages.

Histograms - x axis is bins or ranges, so for example if you had a tempo histogram, there might be a tempo bin from 120 to 125 BPM. All songs that fit that category would stack in that bin. Y axis is the count or frequency in which that bin occurs. This gives you an idea where the user (or multiple users) have most of there tempos (or whatever measurement).

As for all of the indexes, these come from Spotify’s open API. Like I said, I did little analysis this time round. It was more of me learning how to extract all this information from Spotify, and also learning to script this visualization toolkit (plotly). There’s proper theory behind what they’re doing - descriptions can be found here:

https://developer.spotify.com/documentation/web-api/reference/tracks/get-audio-features/

Spotify is conducts layers of analytics on music. This level is called “audio features” which is a child class of their “audio analysis.” The latter is dissecting pitches, pitch changes, timbres, modes, etc., and the former is running some model.

I apologize for maybe being misleading regarding acousticness, liveliness, and instrumentalness. Those actually appear to be measures of confidence that the track is acoustic, live, or instrumental. The rest are proper indexes still. With that being said, I would suspect that yes, the more instrumental or more acoustic a song is, the more it will trigger the measurement. With that also being said, these metrics are a little less relevant now.
 
Back
Top Bottom